KR930020860A - Image compression encoding and decoding method and apparatus - Google Patents

Abstract

In the compression encoding and decoding method and apparatus of the present invention, the DC coefficient of the DCT block is scalar-quantized and the AC coefficient is classified vector quantization (CVQ).

The minimum value of the ground value obtained by subtracting some or all of the horizontal AC coefficient including the first AC coefficient and the vertical AC coefficient including the second AC coefficient by the zigzag scanning order of the DCT block to the representative value of the preset reference class. Classify each DCT block according to Using the hierarchical compressor method, the lowest level codes are separated into P pieces with equal intervals for each classified DCT block, provide a codebook of corresponding representative vectors, and the corresponding representative vectors in each codebook. The indices and the classification codes of are provided to have a constant bit at all times as encoding data corresponding to AC coefficients. In addition, the error generated at the level immediately before is obtained, and the codebooks of the corresponding S representative vectors are provided again, and the multi-stage subclass vector quantization of the schemes for providing the corresponding indexes and the additional information in each codebook is encoded. In this case, the lower level compresses the image, and the higher level outputs the code having the finer image quality.

When decoding, decoding is performed in the reverse order of encoding. However, during fast searching, only the lowest level code having a certain number of bits is restored regardless of the complexity of the screen, so that the image is reproduced without deterioration of image quality even during high-speed search and special playback, and image quality is improved.

Description

Image compression encoding and decoding method and apparatus

Since this is an open matter, no full text was included.

1 is a block diagram of an image compression encoding apparatus according to the present invention.

2 is a view for explaining the DC and AC coefficients of the 8 × 8 DCT block.

3 is a view for explaining class classification of an image compression encoding apparatus according to FIG. 1;

Claims (13)

In the compression encoding method for dividing a 2D video signal into discrete cosine transform (DCT) blocks and encoding a transform coefficient of each DCT block at a predetermined compression rate, the DC coefficient of each DCT block is scalar quantized to n bits, The AC coefficients of the DCT block are classified into the m classes, and the AC coefficients of the classified DCT blocks are approximated with the f representative vectors set for each class, and then quantized into the g-bit indexes of each representative vector to encode for each block. Number of bits (B) B = n + log 2 m + f × g Lowest level vector quantization occurring at a constant bit rate of; Each representative vector is approximated by h representative vectors set for each of the classified classes according to the result of the vector quantization in the previous quantization process and the components according to the values of all AC coefficients of the DCT block, that is, the error generated in the previous quantization process. A plurality of hierarchical level vector quantization processes for vector quantizing with an index of g bits of a; Determining a parameter in consideration of the target number of bits allocated to the screen based on the encoding result of the video signal of the previous screen, and adaptively assigning bits to the vectors of the plurality of vector quantization processes to the vector higher than the parameter. An image compression encoding method comprising an internal vector assignment process. The method of claim 1, wherein the class classification is performed in advance to the AC coefficient by using some or all of the AC coefficient in the vertical axis including the first AC coefficient and the AC coefficient in the horizontal axis including the second AC coefficient in the zigzag scan of each DCT block. An edge compression method is detected according to a minimum value of a square value subtracted from a representative value of a set reference class, and each DCT block is classified according to a detection result. The image compression encoding method according to claim 1, wherein the lowest quantization process and the plurality of hierarchical quantization processes perform vector separation by equally spaced sampling. The image compression method according to claim 3, wherein the lowest level vector quantization process performs vector separation by three equally spaced sampling, and the plurality of hierarchical quantization processes perform vector separation by nine equally spaced sampling. Encoding Method. The image compression encoding according to claim 1, wherein each variable of the lowest level vector quantization process is n, m, p, and g are 8, 8, 3, and 7, and the total number of bits per block is 32. Way. The method of claim 1, wherein all representative vectors of the lowest level vector quantization process and the plurality of hierarchical vector quantization processes are rearranged according to statistics, such as activity or variance, and the vector of the vector according to activity or variance of the input vector. An image compression encoding method characterized by shortening the calculation time by setting the search area to a predetermined range of all areas. The video compression encoding method of claim 1, wherein the parameter of the internal vector allocation process is determined according to the comparison result by comparing the image encoding result and the target bit of the previous screen in consideration of activity. The image compression encoding method of claim 1, wherein the internal vector allocation process satisfies a target bit by differential coding based on the weight of each region of the image. 2. The unit of claim 1, wherein the unit to be recorded on the tape is a sync block, which is a sync signal indicating the start and end of a block, an index signal in a frame, and a lowest level carried in an IDC (Indepedent Decodable Code) area. And a coding code of the plurality of hierarchical level vector quantization processes and additional information carried in a decant coded code, additional information, and a dedependent dependent code (DDC) region. A compression encoding apparatus for dividing a two-dimensional image signal into discrete cosine transform (DCT) blocks and encoding a transform coefficient of each DCT block at a predetermined compression rate, the DC coefficient of the DCT block being a scalar quantization; Classification means for classifying the DCT block into m classes according to an edge type AC coefficient; The lowest level quantization means for quantizing the AC coefficients of the DCT block by f representative vectors set for each class classified by the classification means and quantizing the vectors by an index of g bits of each representative vector; The value of the representative vector of the previous level vector quantization means and the values of all AC coefficients of the DCT block, that is, the Eri generated from the previous level encoding, are approximated by h representative vectors for each classified class, so that a plurality of hierarchical level quantization means for vector quantizing with an index of g bits; Based on the encoding result of the video signal of the immediately preceding screen, a parameter considering the correlation with the target number of bits allocated to the screen is determined, and the plurality of level vector quantization means adaptively converts the vector into a bit with respect to the vector higher than the parameter. And an internal vector assignment means for assigning. DC code and AC coefficient for discrete cosine (DCT) blocks are divided into classes according to the directionality of the edges. Codebooks are set for each class, and they are compressed to be lower level codes, and hierarchical with multiple level codes. In the compression decoding method of decoding a video signal compressed using a compression method of a structure to a predetermined elongation rate and performing inverse DCT conversion, a DC code is scalar inverse quantized to be restored to a DC coefficient, and a codebook corresponding to the class code is used. A least significant vector inverse quantization process for decoding into a vector value of a representative vector corresponding to each level of code; A plurality of vector inverseization processes for decoding the vector value of the representative vector corresponding to the code of each level in the codebook corresponding to the class code; Recognizing the truck regeneration / normal regeneration mode, inverse DCT conversion of the result of adding the lowest vector inverseization process and the plurality of vector inverseization process in normal reproduction and inverse DCT conversion of the result of the lowest vector inverse quantization in special reproduction And a third process of converting the image compression encoding method. DC code and AC coefficient for discrete cosine (DCT) blocks are divided into classes according to the directionality of the edges. Codebooks are set for each class, and they are compressed to be lower level codes, and hierarchical with multiple level codes. A compression decoding apparatus for decoding a video signal compressed using a compression method having a predetermined structure and performing inverse DCT conversion, comprising: DC decoding means for reconstructing the DC code into a DC coefficient by scalar inverse quantization; Least significant vector inverse quantization means for decoding into a vector value of a representative vector corresponding to a least significant code in a codebook corresponding to the class code; A plurality of vector inverse generating means for decoding in a codebook corresponding to a class code with a vector value of a representative vector corresponding to each level of code; When the operation state is normal playback, the DC decoding means, the least significant inverse generating means, and the plurality of vector inverse generating means are controlled to inverse DCT conversion, and during the special reproduction, the outputs of the DC decoding means and the least significant vector inverse generating means. And control means for controlling the inverse DCT conversion. 13. The apparatus according to claim 12, wherein said control means comprises: first adding means for adding an output of said DC decoding means and an output of said least significant vector inverse-zeroing means; Second adding means for adding outputs of the plurality of vector inverse magnetizing means; Third adding means for adding an output of said first and second adding means to output inverse DCT conversion; And mode selection means for controlling the output of the second adding means not to be transmitted to the third adding means in the special playback operation according to the normal playback / special playback operation mode according to the user's key input. Decryption device. ※ Note: The disclosure is based on the initial application.